Make clearer that the memory team is working
to reduce the memory requirement

I'm adding some wording to our hardware requirements to explain that needs vary
by workload. I also included some examples of factors that may influence the
workload.

Update the Unicorn worker documentation based on the new scaling
formula.

https://gitlab.com/gitlab-org/omnibus-gitlab/issues/4535

* Use `GiB memory` to describe memory needs.
* Add local storage recommendation for monitoring node.
* Better align CPU/Memory to match common cloud provider instnace types.
* Update requirements based on the new 10k user standard.

Updating redirected links in the development, install, policy
raketasks and university dirs. Should now link to final
URL to avoid redirect hops